getmusicbee.com

General => General Discussions => Topic started by: hiccup on September 11, 2022, 07:23:53 PM

Title: flac 1.4.0 crashtest
Post by: hiccup on September 11, 2022, 07:23:53 PM
I noticed that flac 1.4.0 was released this weekend.
Out of curiosity I checked it out, and noticed they also have released a 'testbench', which consists of 64 tracks that each has its own challenges that could make a player/decoder trip. (freeze, crash etc.)
https://xiph.org/flac/2022/09/08/decoder-testbench.html

I checked them all, and almost all of them played without any problem.
The only exceptions were:

#55 (which combines a lot of 'you shouldn't do this' in one track)
MusicBee won't load it and hangs at 'scanning'.

#59 (which has artwork embedded as AV1)
It plays, but no artwork is displayed.

So with all but one problematic files MusicBee passed with flying colours.

Perhaps something could be improved so that #55 would also play (but I don't think anyone will ever encounter a track having all these issues at the same time), and perhaps it would be good to look at embedded AV1 images being recognised and displayed.

But I think both are pretty much academical and hypothetical issues.
Title: Re: flac 1.4.0 crashtest
Post by: sveakul on September 11, 2022, 09:56:01 PM
Good to know, thanks hiccup!  For those of us inclined (like me) to update their flac.exe to the new 1.4.0, thought I'd pass on that it now requires the presence of an updated libFLAC.dll to operate.  That according to the readme in the developer's package, which includes both files: https://ftp.osuosl.org/pub/xiph/releases/flac/flac-1.4.0-win.zip (https://ftp.osuosl.org/pub/xiph/releases/flac/flac-1.4.0-win.zip) .

The "complete" (vs. incremental) version of MusicBee was released with a libFLAC.dll in the main directory, but that version is not compatible with the new FLAC.  Also, the libFLAC.dll included in the 1.4.0 package must reside in the same directory as flac.exe.

The main change acc. to the website looks like:  "FLAC can now encode and decode 32 bit-per-sample audio... Note that this is 32 bit integer samples, not 32 bit float samples."